You are leaving Medscape Education
Cancel Continue
Log in to save activities Your saved activities will show here so that you can easily access them whenever you're ready. Log in here CME & Education Log in to keep track of your credits.


Antidepressants in Bipolar Depression: Data From STEP-BD: An Expert Interview With Andrew Nierenberg, MD


Editor's Note:

Does adding an antidepressant to mood stabilizer put patients with bipolar depression at risk of mania? A new and comprehensive study from the NIH yields some results that may change the way you prescribe.

Medscape: Recently Gary Sachs, MD, you, and colleagues conducted some research[1] that created a buzz. Could you briefly set the stage for the rationale and need for this study?

Andrew Nierenberg, MD: One of the interesting things you find when you review the literature is the remarkable dearth of studies that carefully and rigorously look at whether antidepressants per se are effective in bipolar depression. Very few studies have ever been done, and many are either small, or short, or they include patients who aren't really representative of the types of patients who would usually be seen in clinical care. So the need was to try to establish: (1) whether antidepressants are helpful if they are added to ongoing treatment with mood stabilizers, and (2) to assess the safety of antidepressants. There is the idea, not supported by a lot of data, that the modern antidepressant might actually exacerbate the course of bipolar disorder, that they could cause either rapid cycling or a switch from depression into mania, hypomania, or a mixed state.

Medscape: Would you also describe what the current state of the art is in terms of the role antidepressants play in bipolar disorder?

Dr. Nierenberg: Many of the guidelines do include antidepressants as part of the treatment. They are consistent also in saying, well, if you are going to use an antidepressant, it's probably best if you avoid using tricyclic antidepressants because there is a robust literature that shows you could make a course of bipolar disorder worse. So that being said, there are actually very little data that looks at the whole modern generation of antidepressants; even with that limited data, there have been a few meta-analyses -- although they're not really meta-analyses because they included so few patients, and so few studies that suggest that, in fact, antidepressants may be helpful.

Medscape: Could you describe the study design?

Dr. Nierenberg: This was an embedded placebo-controlled randomized trial. The study design was unique because it was part of the systematic treatment enhancement program for bipolar disorder. Notice it doesn't say the systematic treatment enhancement study. And it was a program because it was a comprehensive system of care, of recordkeeping, of collaborative care for patients. So the whole idea was that you would have patients across the country in centers of excellence, where there were experts in bipolar disorder who could be relied upon to provide guideline-informed care. So any patient in any of the mood states that are associated with bipolar disorder, ranging from being fine to being quite ill, could come into the study.

Entry into the study was through outpatient settings, but included patients who had been hospitalized. It included patients who were bipolar I, bipolar II, or specified schizoaffective bipolar, or patients who have had psychosis, as well as those who could have the full range of psychiatric and medical comorbid conditions usually excluded in run of the mill studies. We had a large population of over 4000 patients, across 22 sites who were followed over time and treated with a philosophy of collaborative care. The patient was informed of what was going on; the measurement and management of their care was shared with the patients and their families. Many of these patients were invited to come into the study if they met the reasonable inclusion and exclusion criteria.

Now what I mean by reasonable is that for example, the study included bupropion (Wellbutrin) and also paroxetine (Paxil). And if a patient was previously exposed to those medications and had a bad reaction, or wouldn't take it, they would be excluded.

Medscape: Could you describe the randomization strategy, specifically the use of equipoise stratification.

Dr. Nierenberg: Equipoise stratification is very well described in an article by Philip Lavori in Biological Psychiatry,[2] and is actually quite a simple idea. In this particular case, the patient and the doctor would have to be in equipoise, would have to be even about whether or not a patient got an antidepressant, because quite frankly we didn't know if the antidepressants help or hurt. The other thing that the patient and doctor would have to be in equipoise about is which medication they got randomized to. So, for example, you could have a patient who didn't care, and didn't think that it mattered which one they got randomized to. That patient would either get the bupropion or bupropion placebo, or they would get paroxetine or paroxetine placebo. So even though they could get randomized to either one, there might be some patients who had had a bad experience with paroxetine, for example, so they would get randomized to bupropion or bupropion placebo, and similarly, if somebody had had a bad experience with bupropion, they were randomized to paroxetine or paroxetine placebo. This was designed to make entry into the study easy and that's why there were actually 2 medications. We thought that there wouldn't be a difference between those medications so it wouldn't matter.

Medscape: Can you comment on the outcome measure used in the study?

Dr. Nierenberg: If you look at most studies, all you have to do is either reach remission or response. Remission usually means getting completely better, while if you reach response, you're at least 50% better without getting all the way better. What we did was we had a definition of durable recovery, which meant that if somebody got better they had to sustain that for 8 weeks and then they were declared to have a durable recovery. And what's remarkable in this particular study is you're talking about between 23% and 27% of patients who met those criteria during the study. So it's clear that there's room for improvement and it's clear that there are other things that we can do. Again, this study doesn't address the use of quetiapine or lamotrigine or the olanzapine/fluoxetine combinations, so at least we do have those evidence-based alternatives.

Medscape: 4360 patients were enrolled in the STEP BD program but only 10% of patients in the program were recruited into the study. What does that say about the generalizability of these results?

Dr. Nierenberg: It's not so much that such a small proportion of patients went into the study, because actually a large proportion of the patients had a prospectively observed depressive episode. The second part is that there were plenty of people who had been exposed to both of those medications who didn't want to come into the study. So there is a clear description in the manuscript itself of who came in and who didn't come into the study. But I think it's a little bit of an exaggeration to say that it actually limits the generalizability because the study is really designed to maximize generalizability.

Medscape: The article suggests that approximately 24.6% in the antidepressant arm, and 28.9% in the placebo arm, did not receive an adequate dose of a mood stabilizer during the study. Does that reflect on how poorly we treat patients with bipolar disorder?

Dr. Nierenberg: No, actually, if you play by the rules of a randomized clinical trial and you say, you have to have an operationally defined adequate dose, what happens to the people for whom that dose was the maximum dose they could take? So it's not necessarily, oh, we were treating them badly, it was that's what they were taking and they wouldn't take more. That was one part of it. The other part is not so much that they wouldn't take it, it's that they would have intolerable side effects or had a previous experience where they wouldn't take any more so it was important that for those who were randomized, we just looked at those who took the adequate dose that was operationalized.

Medscape: Less than 1% of patients in each group attempted suicide. Can you comment on the issue of suicide in bipolar disorder in the context of antidepressant treatment?

Dr. Nierenberg: Remember, at least in this particular study and that's all I could really comment on in terms of what you're asking, we were not just adding an antidepressant to no treatment. These were patients who were taking inadequate antimanic medication. I think once you get into whether the selective serotonin reuptake inhibitors can exacerbate suicidal thoughts, suicidal behavior, it becomes a much more complicated issue that I don't think we should get into.

Medscape: The Gijsman and colleagues[3] meta-analysis found that antidepressants did have an effect on bipolar depression. How would you reconcile those findings with those of this study?

Dr. Nierenberg: The Gijsman and colleagues meta-analysis looked at a few studies and from that they concluded that antidepressants actually worked. The problem with that meta-analysis is that the largest study included in it was the olanzapine/fluoxetine combination, not fluoxetine alone, and that heavily weighs the entire small meta-analysis toward the results of that study. But what's remarkable is, even in that study, the number of patients who were exposed to the combination of olanzapine and fluoxetine was around 86, and that's it. The other studies were substantially smaller than that. So one thing that's really rather amazing is that the STEP-BD study is the largest placebo-controlled study of antidepressants for bipolar disorder ever published.

Medscape: In terms of switch rates, you found no difference between the 2 groups. However, a certain proportion of each of the 2 groups (38% and 42%) had experienced the switch in the past. So if one assumes that these patients may have an underlying neurobiologic vulnerability, was there any difference in switch rates between these subgroups?

Dr. Nierenberg: I don't know if that subanalyses was done yet. We are working on the mediator and moderator analyses.

Medscape: If the National Institutes of Health were in a position to fund another follow-up study, what would you be thinking of?

Dr. Nierenberg: Well, the National Institute of Mental Health (NIMH) is thinking of looking at another study to evaluate the use of adjunct moderate doses of lithium to really see how that works over the long term. This is an important issue that hasn't been looked at.

Medscape: Based on your reading of the literature and specifically the current study, when would you advocate the use of an antidepressant in bipolar depression?

Dr. Nierenberg: Well, the current study really suggests several things. One is that, there was no apparent benefit to adding an antidepressant to an ongoing adequate dose of a mood stabilizer. It doesn't address who should get an antidepressant or who responds to it or who doesn't. Again, we have to do further analyses to understand that. This study also doesn't suggest that if a patient's doing well on an antidepressant, you should stop it. I would also like to add that we are very grateful to the NIMH for having the courage to fund this really innovative study that informs care. We'll see what happens in the future.

Supported by an independent educational grant from GlaxoSmithKline

  • Print